Mark

PyTorch version: 1.13.0.dev20220810+cu116 Train command: python train.py --train_data data_lmdb_release/training/MJ/MJ_train --valid_data data_lmdb_release/training/MJ/MJ_valid --select_data MJ-MJ --batch_ratio 0.5-0.5 --Transformation TPS --FeatureExtraction ResNet --SequenceModeling BiLSTM --Prediction Attn Full log: ``` Traceback (most recent call...

Animations look uncorrectly on android. When you add any type to alert on android it looks uncorrectly. kivy: 2.0.0 kivymd: master branch sweetlaert: master branch  ``` from kivymd.app import...

kivy==2.0.0 kivymd==0.104.2 master branch sweetalert==0.1.5 When sweetalert any type(success, info, etc.) or any image open with field on_text_validate it looks uncorrectly. .

OCR accuracy is very low.

2

Image:  Flutter vision tesseract:  Pytesseract: | мы ааеоог лоно о четоотео м пол олам а на | курица № 7у9214-003-50301584-2014 : изделие кулинарное из мяса птицы с гарниром...

Discussed with @joecummings Currently, `expected_tokens` is explicitly given in each test separately, even it is pretty always not unique. With current formatting, it makes tests looks really ugly and not...

[WIP] Add multimodal_instruct_dataset, update docs

4

#### Context What is the purpose of this PR? Is it to - [x] add a new feature - [ ] fix a bug - [ ] update tests and/or...

[WIP] Move schedulers to training from modules.

12

#### Context What is the purpose of this PR? Is it to - [ ] add a new feature - [X] fix a bug - [X] update tests and/or documentation...

Similar to #1868

We have passed several iterations speaking about custom contrastive losses in torchtune. The last point in this direction was the deprecation of SimPO #2062 and the prohibition of new custom...